基于ARIMA-GARCH模型研究加密货币市场波动性  

Study of Cryptocurrency Market Volatility Based on the ARIMA-GARCH Model

作　　者：候先琴 

机构地区：[1]贵州大学数学与统计学院,贵州 贵阳

出　　处：《运筹与模糊学》2021年第4期387-399,共13页Operations Research and Fuzziology

摘　　要：近来,加密货币已然成为投资者、从业者和研究人员非常感兴趣的其他金融投资资产。然而,很少有研究集中分析来预测加密货币市场的波动性。在本文中,我们考察五只具有代表性的加密货币收益率序列的分布结构知,序列是有偏的且呈现尖峰厚尾分布的同时,还具有收益率聚集及杠杆效应等特征。通过分析加密货币的数据分布特征,我们最终选用改进后基于滚动时间窗的SGED分布的变参数ARIMA-EGARCH动态预测模型来分析预测加密货币收益率序列的内在规律;同时,通过滚动时间窗来规避过度拟合的问题。结果表明,该模型相对较好地拟合了加密货币收益率的变化规律,且具有较好的预测效果,可为投资者和相关机构人员提供一种较好的预测工具。Recently, cryptocurrencies have become other financial investment assets of great interest to investors, practitioners and researchers. However, few studies have focused on analysis to predict volatility in cryptocurrency markets. In this paper, we examine the distribution structure of five representative cryptocurrency yield sequences showing that they are biased and present a peaked thick-tail distribution, and are also characterized by yield aggregation and leverage effects at the same time. By analyzing the data distribution characteristics of cryptocurrency, we finally chose the modified variable parameter ARIMA-EGARCH dynamic prediction model based on the SGED distribution of the rolling time window to analyze the inherent law of predicting the cryptocurrency yield sequence;at the same time, the problem of overfitting is avoided by rolling the time window. The results show that the model relatively well fits the change law of cryptocurrency yield, and has a good predictive effect, which can provide a better predictive tool for investors and related institutional personnel.

关 键 词：GARCH模型 广义误差分布 偏态分布 参数优化 波动性预测 

分 类 号：F83[经济管理—金融学]